Quantum constraint learning for quantum approximate optimization algorithm

TL;DR

This paper presents a quantum machine learning method to learn constrained mixer Hamiltonians for QAOA, enabling flexible encoding of constraints and improving optimization in NISQ-era quantum devices.

Contribution

It introduces a novel approach to encode constraints in QAOA via learned mixer Hamiltonians, adaptable to various constraints and circuit depths.

Findings

The learned mixer Hamiltonian can encode general constraints.

The method allows control over circuit depth and constraint enforcement accuracy.

Performance evaluated using Wasserstein distance metric.

Abstract

The quantum approximate optimization algorithm (QAOA) is a hybrid quantum-classical variational algorithm that offers the potential to handle combinatorial optimization problems. Introducing constraints in such combinatorial optimization problems poses a significant challenge in the extensions of QAOA to support relevant larger-scale problems. This paper introduces a quantum machine learning approach to learn the mixer Hamiltonian required to hard constrain the search subspace. We show that this method can be used for encoding any general form of constraints. One can directly plug the learnt unitary into the QAOA framework using an adaptable ansatz. This procedure gives the flexibility to control the depth of the circuit at the cost of the accuracy of enforcing the constraint, thus having immediate application in the Noisy Intermediate Scale Quantum (NISQ) era. We also develop an intuitive metric that uses Wasserstein distance to assess the performance of general approximate optimization algorithms with/without constraints. Finally, using this metric, we evaluate the performance of the proposed algorithm.